Proceedings of the Annual Meeting of the Berkeley Linguistics · PROCEEDINGS OF THE FORTY-FIRST ANNUAL MEETING OF THE BERKELEY LINGUISTICS SOCIETY February 7-8, 2015 General Session

UC BerkeleyProceedings of the Annual Meeting of the Berkeley Linguistics Society

TitleNon-canonical Noun Incorporation in Bzhedug Adyghe

Permalinkhttps://escholarship.org/uc/item/1k30m5v1

JournalProceedings of the Annual Meeting of the Berkeley Linguistics Society, 41(41)

ISSN2377-1666

AuthorErshova, Ksenia

Publication Date2015-01-01

DOI10.20354/B4414110014 Peer reviewed

eScholarship.org Powered by the California Digital LibraryUniversity of California

https://escholarship.org/uc/item/1k30m5v1

https://escholarship.org

http://www.cdlib.org/

PROCEEDINGS OF THE FORTY-FIRST ANNUAL MEETING OF THE

BERKELEY LINGUISTICS SOCIETY

February 7-8, 2015

General Session

Special SessionFieldwork Methodology

EditorsAnna E. JurgensenHannah Sande

Spencer LamoureuxKenny BaclawskiAlison Zerbe

Berkeley Linguistics SocietyBerkeley, CA, USA

Berkeley Linguistics SocietyUniversity of California, BerkeleyDepartment of Linguistics1203 Dwinelle HallBerkeley, CA 94720-2650USA

All papers copyright c© 2015 by the Berkeley Linguistics Society, Inc.

All rights reserved.

ISSN: 0363-2946

LCCN: 76-640143

i

Contents

Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vForeword . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii

The No Blur Principle Effects as an Emergent Property of Language SystemsFarrell Ackerman, Robert Malouf . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

Intensification and sociolinguistic variation: a corpus studyAndrea Beltrama . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

Tagalog Sluicing RevisitedLena Borise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

Phonological Opacity in Pendau: a Local Constraint Conjunction AnalysisYan Chen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

Proximal Demonstratives in Predicate NPsRyan B . Doran, Gregory Ward . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

Syntax of generic null objects revisitedVera Dvořák . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

Non-canonical Noun Incorporation in Bzhedug AdygheKsenia Ershova . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99

Perceptual distribution of merging phonemesValerie Freeman . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121

Second Position and “Floating” Clitics in WakhiZuzanna Fuchs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133

Some causative alternations in K’iche’, and a unified syntactic derivationJohn Gluckman . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155

The ‘Whole’ Story of Partitive QuantificationKristen A . Greer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175

A Field Method to Describe Spontaneous Motion Events in JapaneseMiyuki Ishibashi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197

ii

On the Derivation of Relative Clauses in Teotitlán del Valle ZapotecNick Kalivoda, Erik Zyman . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219

Gradability and Mimetic Verbs in Japanese: A Frame-Semantic AccountNaoki Kiyama, Kimi Akita . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245

Exhaustivity, Predication and the Semantics of MovementPeter Klecha, Martina Martinović . . . . . . . . . . . . . . . . . . . . . . . . . . . 267

Reevaluating the Diphthong Mergers in Japono-RyukyuanTyler Lau . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287

Pluractionality and the stative vs . eventive contrast in RanmoJenny Lee . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307

Labial Harmonic Shift in Kazakh: Mapping the Pathways and Motivations for DecayAdam G . McCollum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 329

Reference to situation content in Uyghur auxiliary ‘bolmaq’Andrew McKenzie, Gülnar Eziz, Travis Major . . . . . . . . . . . . . . . . . . . . . 353

Case-Marking in Estonian PseudopartitivesMark Norris . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 371

Discourse Coherence and Relativization in KoreanSang-Hee Park . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 397

Negotiating Lexical Uncertainty and Speaker Expertise with DisjunctionChristopher Potts, Roger Levy . . . . . . . . . . . . . . . . . . . . . . . . . . . . 417

Patterns of Misperception of Arabic ConsonantsChelsea Sanker . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 447

The Imperative Split and the Origin of Switch-Reference Markers in NungonHannah Sarvasy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 473

Asymmetries in Long-Distance QRMisako Tanaka . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 493

The cross-linguistic distribution of sign language parametersRachael Tatman . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 503

iii

Homophony and contrast neutralization in Southern Min tone sandhi circleTsz-Him Tsui . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 515

Cultural Transmission of Self-Concept from Parent to Child in Chinese American FamiliesAya Williams, Stephen Chen, Qing Zhou . . . . . . . . . . . . . . . . . . . . . . . 533

Fruits for Animals: Hunting Avoidance Speech Style in MuruiKatarzyna Izabela Wojtylak . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 545

A Quest for Linguistic Authenticity: Cantonese and Putonghua in Postcolonial Hong KongAndrew D . Wong . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 563

iv

v

Acknowledgments

As the Executive Committee of the 41st Annual Meeting of the Berkeley Linguistics Soci-ety, we would like to express our gratitude to the conference participants, volunteers, sessionchairs, faculty, and staff members for their participation. It was your contributions that madethe conference a success. We are especially grateful for the patience, hard work, and supportof Paula Floro and Belen Flores, without whom BLS 41 would not have been possible. Wewould also like to thank the following departments and organizations of the University ofCalifornia, Berkeley for their generous financial support:

Department of LinguisticsStudent Opportunity FundGraduate AssemblyDepartment of PhilosophyDepartment of PsychologyDepartment of Ethnic StudiesFrench DepartmentDepartment of Spanish and PortugueseBerkeley Language CenterOffice of Academic Affairs Vice President

vi

vii

Foreword

This monograph contains a number of the talks given at the 41st Annual Meeting of theBerkeley Linguistics Society, held in Berkeley, California, February 7-8, 2015. The conferenceincluded a General Session and the Special Session Fieldwork Methodology. The 41st AnnualMeeting was planned and run by the second-year graduate students of the Department ofLinguistics at the University of California, Berkeley: Kenny Baclawski, Anna Jurgensen,Spencer Lamoureux, Hannah Sande, and Alison Zerbe.

The original submissions of the papers in this volume were reviewed for style by Anna Jur-gensen and Hannah Sande. Resubmitted papers were edited as necessary by Anna Jurgensenand Kenny Baclawski, and then compiled into the final monograph by Anna Jurgensen. Thefinal monograph was reviewed by Spencer Lamoureux. The endeavor was supported by Ali-son Zerbe’s management of the Berkeley Linguistic Society’s funds for publications.

The BLS 41 Executive CommitteeJuly 2015

viii

99

Non-canonical Noun Incorporation in Bzhedug Adyghe

Ksenia ErshovaUniversity of Chicago∗

1 Introduction

Noun incorporation (NI), i.e. the formation of a new verbal form by combining a verb and anoun into a single phonological word, has been the focus of much discussion among linguists.1

Some researchers argue that it is done in the lexicon (Mithun 1984), others have claimedthat it must be treated as a syntactic process (Sadock 1980; Baker 1988 and subsequentwork). This paper addresses NI in the Bzhedug dialect of Adyghe, a polysynthetic languagefrom the Northwest Caucasian family. Unlike many well-known polysynthetic languages,Adyghe does not have productive verbal NI; however, arguments may be incorporated intothe predicate if it is nominalized (1).

(1) a. hac.e-meguest-pl.obl

s-ja-ze1sg.abs-3pl.io+dat-wait

‘I’m waiting for guests’.

b. sjezes’@K1sg.abs.tire.pst

[hec.’e-je-zen@-m]guest-dat-wait-obl

‘I’m tired of waiting for guests (lit. guest-waiting)’.

One of the widely assumed properties of NI is that it targets only the Patient or Themeof the predicate, both in finite predication (Baker 2009:154) and nominalized constructions(Koptjevskaja-Tamm 1993:99–102). In this paper I present data that appears to challengethis generalization and show that it is in fact a phenomenon separate from what has beencanonically understood to be argument incorporation. NI in Adyghe, while governed by theargument structure of the nominalized predication, is morphosyntactically a nominal process.The restrictions imposed on this process inform us on Adyghe clause structure and the sizeof the nominalized construction. Underlyingly, arguments are hierarchically arranged inaccordance with their level of agentivity, with the more agentive argument positioned higher;this supports the widely established hypothesis that the agent is an argument external to theVP (Kratzer 1996), but is not readily obvious within Adyghe grammar. The incorporation

∗The data was collected in Aul Neshukay (Teuchezhsky District, Republic of Adygea, Russia) in July2014. Examples are in the Bzhedug dialect, unless otherwise noted. I am grateful to the speakers of Adyghefor their generous help, Greg Kobele, Maria Polinsky, Yuri Lander, the audience at BLS’41, and especiallyKarlos Arregi for discussion, comments and criticism. All mistakes are mine.1Abbreviations: abs – absolutive, adv – adverbial, ben – benefactive, caus – causative, com – comitative,dat – dative, dir – directive, erg – ergative, imp – imperative, inf – infinitive, intr – intransitive, io –indirect object, ipf – imperfect, neg – negation, nml – nominalizer, obl – oblique, pl – plural, poss –possessive, pr – possessor, prs – past, refl – reflexive.Transcription notes: c =

>tS, s = S, z = Z, Z =

>dz, Z =

>dZ, s = laminal voiceless fricative, z = laminal voiced

fricative, C. = glottalized consonant, C’ = palatalized consonant.

100

Ksenia Ershova

data provides evidence for a “passive” structure of nominalizations, where the nominalizingmorpheme cuts off the external argument.

The rest of the paper is organized as follows: section 2 provides a short overview ofAdyghe clause structure and the polysynthetic verbal form, as well as general informationon nominalizations; in section 3 I present the incorporation data; section 4 contains myanalysis of nominalized constructions and NI in Adyghe; section 5 concludes.

2 Background on Adyghe

In this section I provide general information on Adyghe clausal and nominal structure,focusing on polysynthesis and case-marking patterns, as well as the basic morphosyntax ofnominalized constructions.

2.1 Polysynthesis

A notable typological trait of Adyghe is prominent polysynthesis, i.e. radical head-marking ofsyntactic relations both in the verbal and nominal domain. All participants of a predicationare indexed on the verb, and a single predicate constitutes a full clause; thus, in (2) we cansee five participants: absolutive first person, ergative third person plural and three indirectobjects introduced by their respective applicative heads.

(2) [s@]-q@-[t-de]-[p-f@]-[∅-r]-[a]-Ka-Zes’t@K[1sg.abs]-dir-[1pl.io-com]-[2sg.io-ben]-[3sg.io-dat]-[3pl.erg]-caus-read.ipf

‘They were making me read it to you together with us’ (Temirgoy; Letuchiy 2015)

Within a nominal phrase, the relation of possession is expressed on the nominal, and thefull NP referring to the possessor is optional, analogous to NPs in a full clause:

(3) w-j@[email protected]‘your house’ (Temirgoy; Gorbunova 2009:147)

Participants are marked with a set of personal markers which are uniform across syntacticroles and phonologically very similar to full pronominals (Table 1). Personal markers notassociated with overt morphology are marked as ∅ in this section for illustrative purposes;they are left unmarked in the rest of the paper.

Indirect object markers are always associated with an overt applicative head; there is overa dozen of them in Adyghe. Some examples of these are the dative (j)e-, general locatives’@-, the benefactive fe-, malefactive s.

we- and comitative de-. A predicate may have multipleapplicative markers, example (2) contains three: the dative, comitative and benefactive.

Morphemes within the polysynthetic word are strictly ordered; a slightly simplified verbaltemplate is shown in Table 2.

The nominal template is virtually identical; it contains all the same zones. In (4) wesee the possessive marker, which belongs to the argument structure zone (A) preceding themarker of negation, a pre-base element (zone B), which in turn precedes all lexical roots(zone C).

101


pronouns abs io / pr erg1sg se s@- s-1pl te t@- t-2sg we w@- w-//p-2pl swe sw@- sw-3sg

demonstratives∅ ∅ @-//j@-

3pl ∅ a-

Table 1: Cross-reference markers and pronouns (Arkadiev et al. 2009:45,56)

Argument structure zone (A) Pre-base elements (B)

ABS DIR APPL DAT ERG PRS / NEG1 2 3 4 5 6

Base (C) Endings (D)

CAUS root APPL TAM PL PRS NEG, case7 8 9 10 11 12 13

Table 2: Simplified verbal template (Arkadiev et al. 2009:42)

(4) [j@]A-[m@]B-[Peń@ne-ńape]Cposs-neg-ring-expensive

as’he.obl

q@s’taK3sg.erg-take.pst

‘He took someone else’s expensive ring’ (Lander 2014)

The strict morphological ordering is especially evident if a nominal is used predicatively.In (5) the causative morpheme is wedged in between the nominal possessive marker andthe lexical root of the nominal, seemingly creating a contradiction between the ordering ofmorphological markers and their semantic and syntactic scope.

(5) [z-j@-z]A-[Ke]B-[s@-nah@c. ’a-K]Crefl.pr-poss-1sg.erg-caus-brother-younger-pst‘I made him my younger brother’. (Lander 2014)

While all participants of a predication are indexed on the verb, none of them may beincorporated into a finite predicate. Nominals, however, incorporate lexical roots: non-referential adjectives and nouns combine with the head noun into a complex word stem, asone can see in (4) and (5).

2.2 Ergativity

As can be seen in Table 2, the verbal personal markers are organized in accordance withergative alignment: the direct object of a transitive verb, as well as the intransitive subject, iscross-referenced in the absolutive slot (slot 1 in Table 2), while the ergative is marked closerto the root, in slot 5.2 Ergativity manifests itself in case-marking as well: the intransitive

2Personal markers in the polysynthetic verbal form display some traits which liken them to pronominalclitics, as proposed for such languages by Jelinek (1984). This is not directly relevant to the topic addressedin this paper; here I avoid the terms ‘agreement’ or ‘cliticization’ for the purposes of neutrality.

102

Ksenia Ershova

subject (6a) and the transitive direct object (6b) are marked with the absolutive case -r,while the transitive subject is marked with -m (6b).

(6) a. c. ’ale-rboy-abs

∅-qese3sg.abs-dance.prs

‘The boy is dancing’.

b. zwak.we-m

plowman-oblqw@bKwe-rfield-abs

∅[email protected]

‘The plowman plowed the field’. (Temirgoy, Arkadiev et al. 2009:53)

The marker -m is glossed as oblique, rather than ergative, because it is not restrictedto the ergative argument: it also marks all indirect objects introduced by applicatives (thecomitative sw@z@ ‘woman’ in (7)), possessors (8) and complements of postpositions (9).3

(7) ń.@-rman-abs

sw@z@-mwoman-obl

∅-[∅-d][email protected][3sg.io-com]-help.prs

‘The husband is helping the wife’.(Temirgoy; Arkadiev et al. 2009:53)

(8) s.en@Ke-mknowledge-obl

∅[email protected]

‘Knowledge Day (September 1)’

(9) hac.’e-meguest-pl.obl

apaje3pl.for

‘for the guests’

Both the absolutive and the oblique case markers may be dropped; overt case morphologycorrelates with definiteness/referentiality. Proper names, first and second person pronounsand possessed NPs are not marked with case (Arkadiev et al. 2009:51-52).

Several authors have argued that morphological ergativity correlates with syntactic erga-tivity, i.e. for a structure within which the ergative is lower than the absolutive (Landerto appear; Letuchiy 2010). The data presented in this paper does not directly challengesuch claims, but presents evidence for a higher position of the ergative subject on the levelof base-generation, supporting the underlying clause structure proposed by Caponigro andPolinsky (2011) for Adyghe.

2.3 Defining the word in Adyghe

Nominalization in Adyghe involves argument incorporation, i.e. the formation of a complexword by combining the nominalized verb and one of its arguments. To define incorporationin Adyghe one must first determine the markers of a word boundary. This section outlinesthe main diagnostics for defining the word in Adyghe, as provided in (Lander 2012).

There are two main parameters which allow to determine word boundaries:

1. Strict morphological organization of the word form (see section 2.1).

3The plural is marked by a separate morpheme -xe, but in the oblique the combination -xe-m ‘pl-obl’ canbe optionally replaced by a portmanteau morpheme -me, as in (9).

103


2. Phonological alternation in penultimate syllable of the stem, i.e. the full word, exclud-ing endings (zone D in Table 2):

(10) /e/ → /a/ | Ce]CIf a foot contains two syllables of the form Ce, where C stands for one consonantor a two-consonant cluster, and is located at the right edge of the stem (zoneC), the vowel in the first of the two syllables changes from /e/ to /a/ (Arkadievet al. 2009:29)

This can be seen in the following examples. In (11a) the last (and only) two syllablesof the stem (zone C) are of the form Ce, and thus the vowel in the penultimate syllablebecomes /a/. In (11b), on the other hand, while the stem contains the sequence of the formCeCe, it is not located at the right edge of the stem, and thus no alternation takes place.

(11) a. [beZe]C-[r]D > baZe-rfox-abs‘fox’

b. [beZe-b@]C > beZe-b@fox-hole‘fox hole’ (Temirgoy, Arkadiev et al. 2009:29)

In (11b) the root beZe ‘fox’ is an incorporated modifier of the root b@ ‘hole’. Incorporationin nominalized constructions is largely analogous, as we will see in section 3.

In the following subsection I provide the basic information regarding the morphosyntaxof nominalized constructions.

2.4 Nominalizations

This paper focuses on two types of nominalized constructions: the action nominal markedwith the suffix -n (12) and the manner nominal marked with the suffix -c. ’e (13). Whilesemantically different, these constructions appear to exhibit identical morphosyntactic be-havior, hence I use examples of both interchangeably and gloss both uniformly as ‘nml’.

(12) [hac.’e-xe-mguest-pl-obl

ja-je-ze-n]3pl.poss-dat-wait-nml

ZarineZarina

[email protected]

‘Waiting for guests is Zarina’s task’.

(13) [w-j@-therqwe-Keska-c. ’e]2sg.pr-poss-pigeon-feed-nml

s@[email protected]

jeKape3sg.erg-worry.prs

‘Your manner of feeding pigeons irritates me’.

These constructions, while derived from predicates, have largely nominal morphosyn-tax. Absolutive and ergative personal markers are necessarily dropped, and indirect objectmarkers are generally dropped as well. The nominalized verbs are modified by incorpo-rated adjectives, as regular nominals, (14a) and, unlike regular predicates (14b), may not bemodified by adverbs (14c).

104

Ksenia Ershova

(14) a. seI

sj@c. ’as1sg.like.prs

[ps@nc.’e-ce-n@-r]fast-run-nml-abs

‘I like to run fast’.

b. seI

sj@c. ’as1sg.like.prs

[ps@c. ’-ewfast-adv

s@-cenew]1sg.abs-run.inf

‘I like to run fast’.

c. * seI

[email protected]. ’as1sg.like.prs

[ps@c. ’-ewfast-adv

ce-n@-r]run-nml-abs

Expected: ‘I like to run fast’.

Nominalizations are not limited to sentential complement positions; they may surface inany nominal syntactic position, e.g. as the complement of a postposition:

(15) seI

stoltable

>ts˙@.Ke

1sg.erg.do.pst[leKe-Kw@s’@z’@-n@-mdish-dry-nml-obl

paj]for

‘I set the table for dish-drying’.

Like regular nominals, they may head a relative clause (16) and appear with demonstra-tives (17).

(16) [qeswa-c.’-ewdance-nml-adv

s-a-Kes.aKe-m-re]1sg.abs-3pl.erg-teach.pst-obl-and

seI

>ts˙et@Ke-m-re

1sg.erg+know.ipf-obl-andńesewvery

[email protected]

‘The manner of dancing which you were taught and the one I know are very different’.

(17) [m@this

w-j@-beZe-w@c. ’@-n]2sg.pr-poss-fly-kill-nml

zec.’[email protected]

jezes’@Ktire.pst

‘Your killing of flies has annoyed everyone’.

The arguments of the nominalized predicates are remapped to positions appropriate fornominal modifiers: as incorporees or possessors. (18a) shows that the absolutive argumentmay be expressed as an incorporee, it may not retain absolutive case marking as in a finiteclause (18b).

(18) a. [leKe-thac.’@-n@-m]dish-wash-nml-obl

s-jezes’@K1sg.abs-tire.pst

‘I’m tired of dish-washing’.

b. * [laKe-xe-rdish-pl-abs

thac.’@-n@-m]wash-nml-obl


Expected: ‘I’m tired of dish-washing’.

A more detailed description of argument encoding, and particularly incorporation, isprovided in the following section.

105


3 Noun incorporation

This section focuses on the morphosyntactic properties of argument incorporation in nom-inalized constructions. I show that this is indeed a case of incorporation, i.e. formationof a new word by combining a nominalized predicate with an argument, and continue todemonstrate that it is a nominal process, rather than verbal. I then offer data that showsthat this phenomenon is nevertheless restricted by verbal argument structure.

3.1 Diagnostics for incorporation

In section 2.3 I outline the main diagnostics for determining word boundaries in Adyghe. Thissubsection is aimed to display that in nominalized constructions the incorporated argumentforms a single word with the predicate based on these diagnostics.

Firstly, the incorporated nominal does not form its own stem for the phonological al-ternation presented in (10). (19a) shows that the penultimate syllable in the root sekwe‘hunt’ surfaces as /a/ in the right environment, i.e. when this root functions as an indepen-dent word; in (19b), where the corresponding nominal is incorporated into the nominalizedpredicate, the alternation no longer takes place.

(19) a. seI

s@gw@ [email protected]

[sak.we]

hunts@-k.

wenew1sg.abs-go.inf

(< sek.we)

‘I like to go hunting’.

b. seI


[sek.we-k.we-n@-r]

hunt-go-nml-abs

‘I like to go hunting’.

Secondly, the incorporated root can be embedded in morphology relating to the fullnominalized form. In (20) the second person possessive marker preceding the incorporatedroot ńeKe ‘dish’ refers to the full nominalized form, and not just to the nominal directly tothe right; this is evident from the English translation.

(20) s’@Ket@z’stop.imp

[w-j@-[leKe-a-fe-thac.’@-c. ’e]]2sg.pr-poss-dish-3pl.io-ben-wash-nml

‘Stop washing other people’s dishes!’

The morphosyntactic position of the root leKe ‘dish’ indicates that it forms a singlemorphophonological unit with the nominalized predicate.

In the following subsection I show that NI in Adyghe, if treated as a case of argument in-corporation in Baker’s (1988) sense, challenges the generalization that this type of operationis only possible for the direct complement of the lexical verb, and consequently, existing ac-counts of this operation. In subsection 3.3 I proceed to show, however, that this phenomenonis not in fact the same, but can be explained as a nominal process.

106

Ksenia Ershova

3.2 Unrestricted noun incorporation

A widely documented cross-linguistic property of noun incorporation is that it only targetsthe Theme or direct object of the verb (Baker 2009:154). This has been claimed to be true fornominalized predicates as well (Koptjevskaja-Tamm 1993:99–102). Incorporation in Adyghenominalizations appears to challenge this generalization.

In nominalized constructions, any argument may be incorporated: the direct object(20),the unaccusative (21) or unergative (28) subject, an applicative indirect object (22) and eventhe ergative subject (23), (24).4

(21) seI

stoltable

>ts˙@Ke

1sg.erg.do.pst[leKe-Kw@s’@z’@-n@-m[dish]SUBJ-dry.intr-nml-obl

paj]for

‘I set the table for dish-drying’.

(22) seI


[Pweshe-tje-t@-n@-r][hill]IO-loc-stand-nml-abs

‘I like standing on a hill’.

(23) [beZ’e-z@-thac.’@-c. ’e-r][fox]SUBJ-[refl.abs]OBJ-wash-nml-abs

c. ’et@w-z@-thac.’e-mcat-refl.abs-wash-obl

fedlike

‘The fox’s manner of washing itself is similar to the cat’s’.

Furthemore, some speakers allow incorporation of more than one argument; in (24) theabsolutive and ergative arguments are both expressed as incorporees.

(24) [psese-leKe-thac.@-c. ’e-r][girl]SUBJ-[dish]OBJ-wash-nml-abs


‘I like the girls’ way of washing the dishes (as if there’s a girls’ type of dish-washing)’.

Some examples even seem to have a possessor and an incorporee referring to the samesyntactic argument. Thus, in (25) and (26), both the incorporee and possessive phrase seemto represent the absolutive subject of the nominalized verb, and the ergative subject in (27).

(25) [w-j@-beZ’e-s’x@pc.@-c. ’e][2sg.pr-poss]SUBJ-[fox]SUBJ-smile-nml

[email protected]@.r.ep1sg.abs-3sg.erg-calm.prs.neg

‘Your fox-like grin disturbs me’.

(26) [psase-m[girl-obl]SUBJ

j@-c. ’ele-qe.swa-c.’e]

poss-[boy]SUBJ-dance-nmlKes.eK

weninteresting

‘It’s interesting when the girl does a male dance’.

4In this and following sections I label incorporated arguments and possessors with their correspondingsyntactic roles in a finite clause, using the abbreviations subj for intransitive subjects and transitive agents,io for indirect objects and obj for direct objects. I use these terms for illustrative purposes solely, todistinguish these labels from morphological case in the nominalized constructions. The term ‘subject’ isparticularly problematic for Adyghe, since it has been proposed that this language is syntactically ergativeand the absolutive displays subjecthood properties (Letuchiy 2010; Lander to appear).

107


(27) [a[that

c. ’ale-mboy-obl]SUBJ

j@-psese-Zene-d@-c. ’e]poss-[girl]SUBJ-[dress]OBJ-sew-nml

senmanner

zec.’j@all

jeKes’[email protected]

‘That boy’s girlish manner of sewing dresses makes everyone laugh’.

I claim that these peculiarly unrestricted patterns of incorporation do not necessarily con-stitute a counterexample to Baker (2009) and Koptjevskaja-Tamm’s (1993) generalization.In the following section I argue that, while an argument is incorporated into the nominalizedpredicate in the sense that the two form a single morphophonological unit, this is not thesame type of mechanism as described by Baker (1988), i.e. it does not involve Head Move-ment of the closest complement of the verb to its verbal head. The strange properties of thisconstruction can be readily accounted for if it is addressed as a relation between a nominaland its syntactic dependents.

3.3 Noun incorporation is nominal

Several empirical facts appear to indicate that incorporation in nominalized constructions isa nominal, rather than verbal process.

One piece of evidence is the morphosyntactic position of the incorporated nominal: itattaches to the left of all verbal morphology, as opposed to next to the verbal stem or inthe position where the corresponding personal marker would surface in a finite construction(Table 3).

POSSincorporatedargument(s)

verbalprefixes

verb rootverbalsuffixes

nomin-alizer

nominalsuffixes &endings

Table 3: Morphosyntactic position of incorporated noun

Thus, we can see in (28) that the nominal ad@Ge-psese ‘Adyghe girl’ appears to theleft of the directive prefix, which, within the verbal template, can be preceded only by theabsolutive personal marker.

(28) seI


[[ad@Ge-psese]-[qe]2-swe-n@-r][Adyghe-girl]ABS-dir-dance-nml-abs

‘I like how Adyghe girls dance’.

Another argument for the nominal nature of argument incorporation in nominalizations isits striking similarities to incorporation of modifiers and nominal arguments into non-derivednouns. In (29) we can see that regular nominals may have several incorporees; for relationalnouns the closest incorporee is interpreted as the argument or inherent possessor of thatnominal (30), while the outer incorporee is interpreted freely. Multiple NI in nominalizedconstructions appears to follow the same pattern: in (24) the inner incorporee is interpretedas the patient, while the outer incorporee denotes a generic agent.

108

Ksenia Ershova

(29) [ad@ge-hec.’ek.wec.’e-sw@ret-xe-r]

Adyghe-[animal]-picture-pl-absqj@[email protected]

‘Adyghe pictures of animals are hard’ (Lander 2014)

(30) j-ane-sposs-[mother]-brother‘his uncle (mother’s brother)’ (Temirgoy; Lander 2015)

As expected for regular nominals, possessive phrases in nominalizations are interpretedfreely, according to context: they can refer to any of the participants, including the directobject (31), or not an argument at all (32).

(31) [laKe-me[dish-pl.obl]OBJ

ja-thac.@-n]3pl.pr+poss-wash-nml

seI

s-jezes’@K1sg.abs-annoy.pst

‘I’m tired of washing the dishes’.

(32) [s-j@-c. ’ale-me[1sg-poss-boy-pl.obl]POSS

ja-pse-thac.@-n]3pl.pr+poss-[neck]OBJ-wash-nml

mafe-qesday-every

s@-p@ń1sg.abs-try

‘Every day I wash the necks of my boys’.

In the following section I show that, as in relational nouns, the closest incorporee in nom-inalizations is structurally restricted, while any additional incorporees are freely interpretednominal modifiers. Crucially, the data indicates that the choice of the closest incorporee isgoverned by underlying verbal structure, providing insights to how much verbal structure isencompassed by the nominalizer, and how much is omitted.

3.4 Incorporation driven by argument structureWhile the previous sections might have created an illusion of chaos in the realm of argument incorporation, there is in fact a restriction regarding this process, which is summarized in(33).

(33) Incorporation Hierarchy: In a two-place predicate, the closest incorporee mustbe the less agentive argument.

This generalization holds for all types of two-place predicates: transitive with an erga-tive agent and absolutive patient (34), intransitive with an absolutive subject and indirectobject introduced by an applicative (35) and so-called inverse verbs with the more agentiveargument introduced by an applicative (36). Below I review each type of verb separately.

1. ERG > ABS: if a transitive predicate is nominalized, the absolutive must be incorpo-rated first.

The predicate thac. ’@- ‘wash’ is an example of a transitive two-place predicate: in (34a)it is used in a finite clause; the direct object laKexer ‘dishes’ is in the absolutive, and thefirst person prefix on the predicate marks the ergative subject. In case this predicate is

109


nominalized, the absolutive argument must be the incorporee closer to the nominalizedpredicate. This can be seen in (34b), where the noun leKe ‘dish’ is now incorporatedinto the predicate, and the ergative agent is expressed as a possessor; (34c) shows thatthe arguments may not be switched in position without a change in meaning, in thiscase rendering a nonsensical utterance.

(34) a. seI

deKw@dedewvery good

laKe-xe-rdish-pl-abs

s-thac.’@Ke-x1sg.erg-wash.pst-pl.abs

‘I washed the dishes thoroughly’.

b. [psase-m[girl-obl]SUBJ

j@-leKe-thac.’@-n]poss-[dish]OBJ-wash-nml


‘I’m tired of the girl’s dish-washing’.

c. #[laKe-me[dish-pl.obl]OBJ

ja-psese-thac.’@-n]3pl.pr+poss-[girl]SUBJ-wash-nml


Expected: ‘I’m tired of the girl’s dish-washing’. (“Seems as if the dishesare washing the girl”.)

2. ABS > APPL: if an intransitive predicate with an indirect object is nominalized, theindirect object must be incorporated first.

An example of such an intransitive two-place predicate is [email protected] ‘congratulate’; in(35a) the one who carries out the action is marked in the absolutive and the addresseeis marked with the oblique and introduced by the benefactive prefix on the predicate.In (35b) we can see that the incorporee that appears closer to the nominalized predicateis necessarily interpreted as the benefactive, i.e. the addressee.

(35) a. sabij-xe-rchild-pl-abs

c. ’elejeKaZe-mteacher-obl

[email protected]

ben-congratulate.pst

‘The children congratulated the teacher’.

b. [psase-me[girl-obl]SUBJ

ja-c.’[email protected].’e]

3pl.pr+poss-[boy]IO-ben-be.happy-nmls@gw@ [email protected]

‘I like how the girls congratulate the boys’.*‘I like how the boys congratulate the girls’.

3. APPL > ABS: if a two-place predicate that has an absolutive and applicative argu-ment, where the applicative argument is more agentive, is nominalized, the absolutiveargument must appear as the closest incorporee.

Inverse verbs in Adyghe constitute a small set of predicates, where the more agentiveargument is introduced by an applicative prefix and carries oblique case, while theless agentive argument is marked as the absolutive (Arkadiev et al. 2009:64–65). Onesuch predicate is s’@Kw@pse- ‘forget’: the experiencer is introduced by the locativeapplicative prefix and the stimulus is marked with the absolutive (36a). If this predicate

110

Ksenia Ershova

is nominalized, the closest incorporee is necessarily interpreted as the absolutive, i.e.the less agentive argument (36b).5

(36) a. c. ’ale-mboy-obl

gw@s’aPe-xe-rword-pl-abs

zec.’eall

s’@[email protected]

‘The boy forgot all the words’.

b. [c.ale-me[boy-pl.obl]EXP

ja-nene-s’@-Kw@psa-c.’e]3pl.pr+poss-[grandmother]TH-loc-forget-nml


‘I don’t like how children forget their grandmothers’.*‘I don’t like how grandmothers forget their children’.

The pattern of argument encoding displayed above for inverse predicates shows that thehierarchy in (33) is based on agentivity, and not surface argument encoding. Intransitivepredicates with an indirect object, as in (35), and inverse predicates, as in (36), have identicalargument structures on the surface: one absolutive and one applicative argument. However,we see that the restrictions on incorporation differ based on the agentivity of the participants.This means that the hierarchy is sensitive not to case-marking or surface positions, but tothe semantic nature of the arguments.

When a two-place predicate is nominalized, one of the arguments can be dropped, includ-ing the less agentive one (37a). An additional structural restriction applies in these cases,however: the more agentive agrument is necessarily expressed as a possessor and cannotbe incorporated; if incorporated, it coerces the interpretation of the less agentive argument(37b).

(37) a. ńesewvery

dej@bad

[nane-me[grandmother-pl.obl]EXP

ja-s’@-Kw@pse-n]3pl.pr+poss-loc-forget-nml

‘It’s very bad, forgetfulness of grandmothers’.

b. ńesewvery

dej@bad

[nene-s’@-Kw@pse-n@-r]grandmotherTH-loc-forget-nml-abs

‘It’s very bad to forget grandmothers’.*‘It’s very bad, forgetfulness of grandmothers’.

Along with the Incorporation Hierarchy summarized in (33), this data indicates thatargument encoding in nominalizations is structurally constrained, and the constrains concernthe closest incorporee, i.e. the less agentive argument. This can be accounted for in astraightforward way if we assume that the verbal argument structure is the driving forcebehind the restrictions. The following section outlines the analysis.

5Since it is particularly hard to assign the term ‘subject’ or ‘object’ to the arguments of this verb, I markthe applicative argument as exp for Experiencer and the absolutive argument as th for Theme.

111


NP

N

brother

NP

mother

NP

NMLVP

V

wash

NP

dish

Inherent possession in NP Nominalized VP

Figure 1: Parallel strutures of relational nouns and nominalizations

4 Analysis

This section aims to provide a unified analysis of the incorporation data presented in theprevious section. As shown in section 3, NI in Adyghe nominalizations is, on the one hand,nominal, and on the other hand, restricted by verbal structure. In two-place predicates,the less agentive argument must appear as the closest incorporee, while the more agentiveargument attaches as a possessor or outer incorporee. In such cases, the outer incorporeeis interpreted as a generic agent (24) or an adjective of manner (27). The incorporee ofunergative one-place predicates, such as s’x@pc@- ‘smile’ and qeswe- ‘dance’ receives an inter-pretation similar to the ergative incorporee: manner (25), (26) or generic agent (28).Thus,we see that the internal argument (direct object of transitive verbs and indirect object ofintransitive verbs) is morphosyntactically restricted analogous to inherent possessors in un-derived nominals, while the external argument – ergative, absolutive or applicative in thecase of inverse predicates – is interpreted analogous to general nominal modifiers. Thisasymmetry points toward a structure of nominalizations which contains only the internalargument and cuts off the external one, analogous to passive nominalizations described byAlexiadou (2001).

Figure 1 presents the simplified structure of the nominalized predicate from example (24)and parallels it to relational nouns which have an internal argument – the inherent possessor(30.6

The proposed structure provides evidence for the ergative and unergative agent, as wellas the applicative in inverse predicates, being an argument external to VP. Furthermore,this argument must be structurally higher than indirect objects introduced by applicativeprefixes, since, as we saw in section 3, these arguments are subject to the same structuralrestrictions as the absolutive direct object. This means that applicative arguments, unlike theagent, must be part of the nominalized structure. Figure 2 contains the basic clause structurefor a transitive three-place predicate, where the part of the verbal structure excluded fromthe nominalization is colored gray.7

6I label the incorporated elements NP since they cannot include demonstratives, but may include adjectivalmodifiers or other incorporated elements.

7Working off the verbal template in Table 2, one might notice that morphology relating to the argumentstructure is prefixal, while temporal and aspectual morphemes attach as suffixes. Here, I assume that thiscorrelates with the corresponding functional heads being left- or right- branching. The language has been

112

Ksenia Ershova

vP

v′

ApplP

Appl’

VP

VDPOBJ

Appl

DPIO

v

DPSUBJ

Figure 2: Basic transitive clause structure

ApplP

Appl′

vP

VP

VDPTH

v

Appl

DPEXP

Figure 3: Structure of inverse predicates

In Figure 2 the only part of the tree that is shaded gray is the external argument.Since nominalized constructions can include overt realizations of v, such as the causativemorpheme, I assume that the nominalizer behaves as a valency-reducing operator, ratherthan cutting v off completely.

The structure for two-place predicates with an absolutive subject and indirect objectare virtually identical to Figure 2, except that V lacks a complement. Two-place inversepredicates include an applicative head above v, which strips the projection beneath it ofits power to license the external argument, the same way as the nominalizer itself, andintroduces an applicative (Figure 3).

Structurally, inverse predicates resemble passives, where a projection above vP stripsthe agentive head of its licensing power. In fact, a number of such verbs are the result ofan agent-demoting operation. For example, the benefactive applicative prefix may carryhabilitive semantics in a construction within which the ergative argument is deleted andreplaced by an indirect object in the benefactive position. In (38a) the predicate ńeKw@- ‘see’

typologically labeled as left-branching, i.e. verb-final, so I assume a projection is right-headed, unless thereis evidence to the contrary. Thus, in Figure 2, v and Appl appear to the left, since they are realized asprefixes, while VP is right-headed.

113


N

NApplP

vP

VP

VDPOBJ

v

Appl

Figure 4: Nominalization of inverse predicate

has a third person plural ergative argument; in (38b), this marker is deleted and replacedby a benefactive, which in turn introduces the external argument.

(38) a. c. ’ale-xe-mboy-pl-obl

bukva-xe-rletter-pl-abs

∅-a-ń[email protected]

‘The boys do not see the letters’.

b. c. ’ale-xe-mboy-pl-obl

bukva-xe-rletter-pl-abs

∅-[a-fe]-ń[email protected][3pl.io-ben]-see-pl.abs-prs-neg

‘The boys cannot see the letters’. (Temirgoy; Letuchiy 2010)

When an inverse predicate is nominalized, the nominalizing projection selects for thehigher ApplP, stripping Appl of the ability to license a specifier. The only argument thatremains within the nominalization is the complement of V, which would have been markedas the absolutive in a finite clause (Figure 4). This is the argument that surfaces as theclosest incorporee in the nominalization.8

The position of the nominalizing projection is further supported by restrictions on personmarking in nominalized predicates: verbal indexing of the absolutive and ergative argumentsis disallowed, while indirect object marking in front of the corresponding applicative is some-times possible: for example, first person singular benefactive in (39) and third person pluraldative in (40).

(39) s@gw@ [email protected]

[w-j@-wered-q@-[s-fe]-Pwa-c.’e]2sg.pr-poss-song-dir-[1sg.io-ben]-say-nml

‘I like how you sing for me’.

(40) s-j-eKegw@mec.’@1sg.abs-3sg.erg-worry.prs

q@weń[email protected]

pepcevery

[w-j@-aXs’e-[ja]-t@-c’e]2sg.pr-poss-money-[3pl.io+dat]-give-nml

‘I am worried about your manner of giving money to everyone who asks you’.

8The nominalizer attaches as a suffix and it thus mapped onto the tree as left-branching head.

114

Ksenia Ershova

NP

N

NnominalizervP

VP

VNPOBJ

v

NPincorporee

Figure 5: Movement analysis of incorporation

This can be accounted for if we assume an analysis within which the ergative and ap-plicative arguments are licensed by their respective heads, and the absolutive is licensed byT. High licensing of the absolutive is supported by its linear position within the verbal form– on the leftmost edge. The nominalizing projection, as agent-demoting applicatives, stripsv of its licensing power, and T is altogether cut off by the nominalizer, while the applicativehead, on the other hand, remains within the nominalization.

The argument structure proposed here (with the ergative positioned higher than theabsolutive) is supported by reflexive binding facts: the ergative binds the absolutive and notvice versa. In Adyghe, reflexivization is achieved by inserting a special reflexive morphemein the morphological position of one of the coreferent arguments in the verb; in (41) wesee that when the transitive predicate ńeKw@- ‘see’ is reflexivized, the reflexive morphemesurfaces in the absolutive slot, while the ergative retains original personal marking.

(41) zec.’eall

c˙@f-xe-rman-pl-abs

z-a-ńeKw@-z’@refl.abs-3pl.erg-see-re

‘All the people see themselves’. (ERG > ABS) (Letuchiy 2010)

Thus, this analysis connects with other aspects of Adyghe morphosyntax.Finally, we must address the nature of noun incorporation in Adyghe. Arguably, polysyn-

thesis involves the spell-out of complex clause-level entities as a single morphophonologicalunit. Within the nominal domain, the requirement of single-word spell-out applies at thelevel of NP, and all lexical and non-lexical entities within this projection merge on the sur-face into a single word. Nominalizations, being a nominal structure, are subject to the samerequirement.

A difficulty we need to account for is the position of the incorporated argument: it surfacesto the left of all verbal morphology, which means that it either moves out of its base-generatedposition within VP (Figure 5) or it merges outside of the nominalized projection and bindsa null pronominal (PRO) within the verbal structure (Figure 6).

It is clear that the overt NP denoting the incorporated argument must be merged outsideof the nominalized projection, but it is not clear which analysis is preferable. The analy-

115


NP

N

NvP

VP

VPROi

v

NPi

Figure 6: Analysis of incorporation with bound PRO

NP

NP

N

N

nml

vP

VP

V

wash

NP

PROi/ti

v

NP

dishi

NP

girl

Figure 7: Structure of nominalized transitive predicate with outer incorporee

sis proposed in Figure 6 has been previously proposed for arguments within nominalizedconstructions (Yoon 1996; Coon 2010), but for subjects, rather than objects.

Within both analyses, outer incorporees and other modifiers attach as adjuncts to NP.Figure 7 reflects the structure of the nominalized predicate in (24), repeated in (42).

(42) psese-[girl]SUBJ-

leKe-[dish]OBJ-

thac.@wash

-c.’e-nml

‘girls’ way of washing dishes’

Possessive phrases attach to the NP and are assigned a loose possessive semantics (“freeR” reading), which is then determined by context (Partee 1996). The most pragmaticallysalient interpretation of the possessor phrase is the unexpressed argument of the nominalizedpredicate; for example, in the nominalized construction from (34), repeated below in (43),the possessor is interpreted as the ergative subject. We know, however, that in the propercontext, the possessor phrase need not denote the external argument (see (31) and (32)).

116

Ksenia Ershova

DP

D′

NP

N

N

nml

vP

VP

V

wash

PROi/ti

v

NP

dishi

D

poss

DP

girl

Figure 8: Structure of nominalized transitive predicate with possessor

(43) psase-m[girl-obl]SUBJ

j@-poss-

leKe-[dish]OBJ-

thac.’@wash

-n-nml

‘the girl’s dish-washing’

Figure 8 provides a rough structure for (43).9

The incorporee immediately adjacent to the nominalizer must be the internal argument;the external argument may not occupy this position (34c), (37b). Neither the movement,nor the binding analysis accounts for this restriction without extra machinery. To enforcethis restriction, I propose that the internal argument (be it a movement trace or a nullpronominal) must be bound by the closest c-commanding nominal phrase within a givendomain, and the domain is NP. This means that in cases like (31), the NP denoting theinternal argument moved out of its incorporated position to the Specifier of DP (Figure 9).This movement was driven by the restriction on referentiality: incorporated elements mustbe non-referential, i.e. no bigger than NP.

In a construction with no overt internal argument, such as (37a), repeated below in (44),the position of the closest incorporee is occupied by a null pronominal (Figure 10).

(44) nane-me[grandmother-pl.obl]EXP

ja-3pl.poss-

∅-[pro]TH

s’@-loc-

Kw@pseforget

-n-nml

‘forgetfulness of grandmothers’

Following this line of reasoning, the external argument interpretation of the incorporatednominal in (37b) is unavailable not for syntactic reason, since the external argument shouldbe able to adjoin as a modifier above the incorporated pro, analogous to the agent in (24).

9For the purposes of this paper, I assume that the possessive morpheme on the nominalization is D; thismight not be the correct analysis for Adyghe, considering that this morpheme has the same morphosyntacticproperties as applicatives and may be present in predicates (5).

117


DP

D′

NP

N

N

nml

vP

VP

V

wash

DP

PROi/ti

v

DP

ti

D

poss

DP

dishes

Figure 9: Movement of closest incorporee to possessor

DP

D′

NP

N

N

nml

ApplP

vP

VP

V

forget

NP

PROi/ti

v

Appl

loc

NP

proi

D

poss

DP

grandmothers

Figure 10: Structure of nominalized two-place predicate with no overt internal argument

118

Ksenia Ershova

I claim that this interpretation is unavailable due to a strong pragmatic bias to match theinternal argument with an overt realization within the domain of NP.

I hope to have demonstrated how the patterns of NI in Adyghe nominalizations can beconnected to the size of the nominalized construction (vP or ApplP in the case of inversepredicates) and general argument structure (the less agentive argument is internal to vP andthus included within the nominalization; the external argument is deleted).

5 Conclusion

Nominalizations in the Bzhedug dialect of Adyghe display a typologically unusual patternof noun incorporation: in these constructions, any argument may be incorporated, includingeven the ergative subject. I show that this morphosyntactic process is not a direct chal-lenge to existing theoretical accounts and is in fact driven by nominal, rather than verbalmorphosyntax. Restrictions on argument incorporation in nominalized constructions pointtowards an argument structure within which the more agentive argument is introduced byan external head; the nominalizer selects for this head and strips it of its licensing power,acting as a valency-reducing operator.

Artemis Alexiadou. Functional Structure in Nominals. John Benjamins, 2001.

P.M. Arkadiev, Y.A. Lander, A.B. Letuchiy, N.R. Sumbatova, and Y.G. Testelets. Vvedenije.osnovnyje svedenija ob adygejskom jazyke [introduction: Basic information on adyghe].In Testelets (2009), pages 17–120.

Mark Baker. Is head movement still needed for noun incorporation? Lingua, 199:148–165,2009.

Mark C. Baker. Incorporation. A Theory of Grammatical Function Changing. University ofChicago Press, Chicago/London, 1988.

Ivano Caponigro and Maria Polinsky. Relative embeddings: A circassian puzzle for thesyntax/semantics interface. Natural Language and Linguistic Theory, 29(1):71–122, 2011.

Jessica Coon. Complementation in Chol (Mayan): A Theory of Split Ergativity. PhD thesis,MIT, 2010.

I.M Gorbunova. Possessivnye konstrukcii [possessive constructions]. In Testelets (2009),pages 146–165.

Eloise Jelinek. Empty categories, case and configurationality. Natural Language and Lin-guistic Theory, 2.1, 1984.

Maria Koptjevskaja-Tamm. Nominalizations. Routledge, London/NY, 1993.

Angelika Kratzer. Severing the external argument from its verb. In Johan Roorych and LaurieZaring, editors, Phrase structure and the lexicon, pages 109–137. Kluwer, Dordrecht, 1996.

119


Yury Lander. Nominal complexes in adyghe: between morphology and syntax. conferencehandout. typology, theory: Caucasus, ifea, istanbul, 1 december 2012, 2012.

Yury Lander. Poluraspad imennogo kompleksa [half-decay of nominal complex]. the adyghelinguistic expedition, neshukay, 18 july 2014, 2014.

Yury Lander. Subject properties of the adyghe absolutive: evidence from relatives. In Volumeon ergativity. Tbilisi State University, to appear.

Alexander Letuchiy. Ergativity in the adyghe system of valency-changing derivations. InGilles Authier and Katharina Haude, editors, Ergativity, Valency and Voice, pages 323–353. De Gruyter Mouton, 2010.

Alexander Letuchiy. Scope versus ordering of operations: causativization and ordering ofvalency-changing operations in adyghe. In Stela Manova, editor, Affix ordering acrosslanguages and frameworks. Oxford University Press, 2015.

Marianne Mithun. The evolution of noun incorporation. Language, 60:847–894, 1984.

Barbara Partee. Genitives: Uniformity vs. versatility: A case study. appendix to theo janssen.compositionality. In Johan van Benthem and Alice ter Meulen, editors, The Handbook ofLogic and Language. Elsevier, Amsterdam, 1996.

Jerrold Sadock. Noun incorporation in greenlandic: A case of syntactic word-formation.Language, 57:300–319, 1980.

Y. G. Testelets, editor. Aspekty polisintetizma: Ocerki po grammatike adygejskogo jazyka[Aspects of Polysynthesis: Sketches on the Grammar of Adyghe]. RGGU, Moscow, 2009.

James H.S. Yoon. Nominal gerund phrases in english as phrasal zero derivations. Linguistics,34:329–356, 1996.

120

Documents

Proceedings of the Annual Meeting of the Berkeley Linguistics · PROCEEDINGS OF THE FORTY-FIRST ANNUAL MEETING OF THE BERKELEY LINGUISTICS SOCIETY February 7-8, 2015 General Session